Simulation design

Simulation.factor EEE.family EEV.family EVV_banana.family EVV_boomerang.family
no contamination EEE EEV EVV_banana EVV_boomerang
noise dimensions correlated EEE_noiseCor EEV_noiseCor EVV_banana_noiseCor EVV_boomerang_noiseCor
rotated (signal in all variables EEE_rot EEV_rot EVV_banana_rot EVV_boomerang_rot

Each simulation is 420 observations of 4 varaibles (3 clusters, with 140 observations each).

Variable space

Principal component space

Radial tours

The starting basis initialized to PC1:2 of each model. Radial tours created for the top 3 variables.

EEE

EEV

EVV_banana

EVV_boomerang

Checking variance-covariance matrices

Let’s check if our component variance-covariance matrices are non-singular and postive semi-definate.

## [1] "cov: cov_circ"
## [1] "is non-singular matrix: TRUE"
## [1] "is positive semi-definite: TRUE"
## [1] "cov: cov_elipse_pos"
## [1] "is non-singular matrix: TRUE"
## [1] "is positive semi-definite: TRUE"
## [1] "cov: cov_elipse_neg"
## [1] "is non-singular matrix: TRUE"
## [1] "is positive semi-definite: TRUE"
## [1] "cov: cov_elipse_pos_cor_noise"
## [1] "is non-singular matrix: TRUE"
## [1] "is positive semi-definite: TRUE"
## [1] "cov: cov_elipse_neg_cor_noise"
## [1] "is non-singular matrix: TRUE"
## [1] "is positive semi-definite: TRUE"

{mclust} paper reference

Scrucca, Luca, Michael Fop, T. Brendan Murphy, and Adrian E. Raftery. “Mclust 5: Clustering, Classification and Density Estimation Using Gaussian Finite Mixture Models.” The R Journal 8, no. 1 (August 2016): 289-317.

mclust paper, Table 3

mclust paper, Figure 2